
The plot method for party and constparty objects are rather flexible and can be extended by panel functions. The pre-defined panel-generating function of class grapcon_generator for pmtrees is documented here.
node_pmterminal(
obj,
coeftable = TRUE,
digits = 2,
confint = TRUE,
plotfun,
nid = function(node) paste0(nam[id_node(node)], ", n = ", node$info$nobs),
...
)
an object of class party.
logical or function. If logical: should a table with
coefficients be added to the plot (TRUE/FALSE)? If function: A function
comparable to coeftable.survreg
.
integer, used for formating numbers.
Should a confidence interval be computed.
Plotting function to be used. Needs to be of format
function(mod, data)
where mod
is the model object.
See examples for more details.
function to retrieve info on what is plottet as node ids.
arguments passed on to plotfun.
# NOT RUN {
if(require("survival")) {
## compute survreg model
mod_surv <- survreg(Surv(futime, fustat) ~ factor(rx), ovarian,
dist = 'weibull')
survreg_plot(mod_surv)
## partition model and plot
tr_surv <- pmtree(mod_surv)
plot(tr_surv, terminal_panel = node_pmterminal(tr_surv, plotfun = survreg_plot,
confint = TRUE))
}
if(require("survival") & require("TH.data")) {
## Load data
data(GBSG2, package = "TH.data")
## Weibull model
bmod <- survreg(Surv(time, cens) ~ horTh, data = GBSG2, model = TRUE)
## Coefficient table
grid.newpage()
coeftable.survreg(bmod)
## partitioned model
tr <- pmtree(bmod)
## plot with specific coeftable
plot(tr, terminal_panel = node_pmterminal(tr, plotfun = survreg_plot,
confint = TRUE, coeftable = coeftable.survreg))
}
# }
Run the code above in your browser using DataLab